Work on speeding up Marpa::R2::HTML

jeffreykegler · Sep 26, 2012 · d28d3ab · d28d3ab
1 parent 310828e
commit d28d3ab
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 13 deletions.
diff --git a/r2/html/lib/Marpa/R2/HTML.pm b/r2/html/lib/Marpa/R2/HTML.pm
@@ -524,8 +524,8 @@ sub handler_find {
 
 say STDERR "handler_find(); action: $action";
 
-        if ( $action =~ / \A SPE_ /xms ) {
-            $handler = $self->{handler_by_species}->{$action};
+        if ( index($action, 'SPE_') == 0 ) {
+            $handler = $self->{handler_by_species}->{substr $action, 4};
             last FIND_HANDLER;
         }
         $class //= q{*};
@@ -899,14 +899,10 @@ sub parse {
         my $marpa_token = $marpa_tokens[$marpa_token_ix];
         last RECCE_RESPONSE if not defined $marpa_token;
 
-        say STDERR "token = ", $marpa_token->[0];
         my $marpa_symbol_id = $grammar->thin_symbol( $marpa_token->[0] );
         my $read_result =
             $recce->alternative( $marpa_symbol_id, PHYSICAL_TOKEN, 1 );
         if ( $read_result != $UNEXPECTED_TOKEN_ID ) {
-            say "UNEXPECTED_TOKEN_ID = ", $UNEXPECTED_TOKEN_ID;
-            say STDERR "result = $read_result ",
-                $LIBMARPA_ERROR_NAMES[$read_result];
             $marpa_token_ix++;
             $recce->earleme_complete();
 	    my $last_html_token_of_marpa_token //= $marpa_token->[1]->[0]->[2];
@@ -1281,10 +1277,19 @@ say STDERR "Start tag token candidate:\n", Data::Dumper::Dumper($start_tag_token
         die "Unexpected step type: $type";
     } ## end STEP: while (1)
 
-    my $value = $stack[0];
+    my $result = $stack[0];
     Marpa::R2::exception('No parse: evaler returned undef')
-        if not defined $value;
-    return \$value;
+        if not defined $result;
+
+    if ( $result->[Marpa::R2::HTML::Internal::TDesc::TYPE] eq 'VALUED_SPAN' )
+    {
+        $result = $result->[Marpa::R2::HTML::Internal::TDesc::VALUE];
+    }
+    else {
+        die "Not yet implemented";
+    }
+
+    return $result;
 
 } ## end sub parse
 

diff --git a/r2/html/lib/Marpa/R2/HTML/Callback.pm b/r2/html/lib/Marpa/R2/HTML/Callback.pm
@@ -190,8 +190,7 @@ sub Marpa::R2::HTML::descendants {
                       $is_valued
                     ? $data->[Marpa::R2::HTML::Internal::TDesc::VALUE]
                     : undef;
-                push @per_descendant_results,
-                    defined $value ? q{} . $value : undef;
+                push @per_descendant_results, $value;
                 next ARGSPEC;
             } ## end if ( $argspec eq 'value' )
             if ( $argspec eq 'original' ) {
@@ -204,7 +203,8 @@ sub Marpa::R2::HTML::descendants {
                     ]
                     )
                     : ( $data, $data );
-                Marpa::R2::HTML::Internal::token_range_to_original(
+                push @per_descendant_results,
+                    Marpa::R2::HTML::Internal::token_range_to_original(
                     $parse_instance, $start_ix, $end_ix );
                 next ARGSPEC;
             } ## end if ( $argspec eq 'original' )

diff --git a/r2/html/script/marpa_r2_html_score b/r2/html/script/marpa_r2_html_score
@@ -108,7 +108,11 @@ my $length_by_element = $instance->{length};
 my $count_by_element = $instance->{count};
 my $html_length = $length_by_element->{html};
 my $total_lengths = List::Util::sum values %{ $length_by_element };
-my $complexity = sprintf "%.3f", ($total_lengths / ($html_length * log ($html_length)));
+my $complexity = 0;
+if ( $html_length >= 1 ) {
+    $complexity = sprintf "%.3f",
+        ( $total_lengths / ( $html_length * log($html_length) ) );
+}
 my $max_depths = $value->{depths};
 my $max_element_depth = $max_depths->{ANY};
 delete $max_depths->{ANY};